That's the point you parasite!

CRISPR isn't random. It's directed by a template RNA strand (called a "guide" and abbreviated sgRNA for historical reasons) to bind sections of DNA complementary to the guide. In addition to matching the guide, the target DNA must have a protospacer adjacent motif (NGG), which limits things a bit in practice.

What happens after target DNA is recognized by the Cas9/sgRNA depends on the specific Cas9 variant and potentially the presence of other exogenous DNA introduced along with the Cas9 and sgRNA.

Gene silencing, or CRISPRi (for interference), targets inactive dCas9 ("dead" Cas9) to the transcription start site (TSS) within a gene. The dCas9 then physically occludes the TSS, preventing transcription factor binding and thus gene expression.

Gene activation, or CRISPRa, also targets the TSS. The dCas9 protein is fused to a constitutively active transcription factor, which initiates gene expression directly.

Editing using CRISPR/Cas9 is also possible. Active Cas9 will make cuts in both strands of the target DNA double-helix, leaving "blunt ends" (unlike the "sticky ends" created by restriction enzymes). The blunt ends are addressed in one of two ways: non-homologous end-joining (NHEJ) or homology-directed repair (HDR). With NHEJ, the blunt ends will either be joined directly (resulting in gene deletion), or with stochastic inclusion of another piece of DNA introduced along with the CRISPR/Cas9 system (resulting in inefficient gene insertion). In HDR, the additional DNA will contain "homology arms" complementary to the DNA flanking the target sequence. After the cuts are made, this DNA can hybridize (forming a double-helix) on those flanking arms. The new DNA finally is copied into the hole created by the Cas9, yielding specific and relatively efficient gene insertion.

Regardless of the specific application, CRISPER/Cas9 approaches are stochastic and imperfect. Off-target DNA can be cut non-specifically, resulting in random deletions or insertion into random regions. Repair may proceed without insertion of the desired sequences, etc. There is a ton of ongoing research into engineered Cas9 variants with improved specificity and efficiency.

Most concerns about CRISPR-based genetic modification relate either to off-target effects (i.e. specificity and/or efficiency), or to the relative ease in making persistent, germ-line changes compared to prior methods.

I think that's exactly right. Look at his statements on trade, e.g. that a country we trade with is "up 100 billion on us" and not that it was a free exchange where they got a financial asset they preferred to their real goods, and we got their real goods which we preferred to our financial assets.

It's partly because of an incredibly simplistic, mercantilist perspective in which trade is intrinsically zero-sum. It's also partly because having never made a dime without bilking somebody, he cannot conceive of someone else agreeing to a fair exchange perceived as a good deal by both sides.

We are comparing relative performance. Uber's vehicles needed hundreds of times more human interventions over the same distance compared to Waymo's.

And if the Tempe police say it, it must be true. /s

The video shows a situation in which a human would have at least tried to react. Uber's car didn't even start slowing, and it has LIDAR. Contrary to what you seem to believe, no current self-driving car technology employs computer vision using a visible light camera to guide the car. Instead, they use IR-band LIDAR, detailed mapping data, and other sensor packages common in existing assisted-navigation / crash avoidance systems (e.g. small RADARs).

Anyway, it's confirmed - Uber's program is far less safe, and less advanced, than Waymo's.

No, it raises further questions about Uber's poor, perhaps criminally negligent, implementation. In the last year Uber's had more, and more serious, accidents than I think every other driverless program combined. Google/Waymo has been testing in San Francisco - not Tempe - for years with nothing comparable.

Me too, and my answer to that interpretation is that there damn well better be. So obnoxious to have to always untar inside a new directory, in case there isn't, and then move the tree up and rmdir if there is, every single time.

The way I wrote that makes it sound like more students enter faculty positions than not, which is incorrect. The national average (US) is ~8%, whereas at my school it's closer to ~35%.

I agree. My university has a much higher proportion of graduates entering faculty positions, yet it also helps students enter suitable industry positions, fosters industry contact, and provides several ways for students to pursue their own entrepreneurship. There's certainly demand for highly trained individuals from all STEM fields across multiple industries.

I'd also add that in my experience, which includes a Master's program at a lower-tier school and a PhD program at an elite one, students are actually pretty realistic about their prospects, and most do not really intend to enter academic careers. They are often willing to give it a shot, or know they'll seek an academic postdoc before entering industry, but the number who have principal investigator as their only desired career outcome is pretty similar to the average proportion of graduates who do attain that status.

My feeling (reinforced through peer communication at conferences, etc.) is that most students know they'll go to industry, but they may not feel safe in openly declaring this to advisers or program directors. No doubt this is a strong confounder for surveys like the one in TFA.

It's allowing them to claim these weren't mass lay-offs, which are separately regulated. See e.g. CA WARN law. There are probably others, as well.

Please. It's clear both incidents were mass lay-offs. In the Solar City division performance reviews hadn't even been carried out.

You and she are mistaken.

They arguably don't use it for editing, just for cutting and degradation to protect against phage. There was recently even published a viral CRISPR-like system, which functions in immunity against viruses that infect other viruses. See "MIMIVIRE is a defence system in mimivirus that confers resistance to virophage."

That's a huge synthesis, we routinely synthesize much smaller pieces. I think it's about a grand for a custom ~5 kbp construct. I guess it's just a matter of time until such large sequences can be synthesized with reasonable time/money investments though.

The zebra fish labs are definitely injecting CRISPR into zygotes to try and create stable edited lines. It doesn't work very well, but it's because zebra fish don't do homology directed repair, and non-homologous end joining is much less efficient. I think the microinjection isn't that bad since the eggs aren't that small. I also know someone who uses a micro needle to poke mammalian cells and - without lysing them - pluck their microtubules in order to measure their viscoelastic properties as they bounce back. I think that's much more difficult.

True, but in practice editing will also require multiple tries, at least for a very long time. There will always be some risk of off-target effects, and it can't be perfectly efficient. People will want to give a few tries and use a validated result.

Whole chromosomes are way, way bigger than any DNA molecules we know how to synthesize right now (by around 5 orders of magnitude). The second step is possible though, we do it with current editing technologies. One of my friends/colleagues even does it to zebra fish zygotes. We've already made it into vertebrates.

Incidentally, zebra fish will also express plasmids, which I thought was completely insane until I learned the injection volume is more than 50% that of the cell. The trick is not to bust 'em open when you do it.